[None][fix] only support deepep post quant all2all on nvfp4 #8041

yilin-void · 2025-09-28T03:29:32Z

Summary by CodeRabbit

Bug Fixes
- Corrected handling of post‑quantization all‑to‑all support across execution methods, ensuring the right path is selected at runtime.
- Added stricter runtime validation to prevent unsupported post‑quantization configurations, reducing unexpected errors.
Refactor
- Centralized the decision logic for post‑quantization all‑to‑all support to improve clarity and consistency in execution paths.
Performance/Stability
- More predictable behavior and safer fallbacks for post‑quantization communication, enhancing overall robustness.

Description

Test Coverage

PR Checklist

Please review the following before submitting your PR:

PR description clearly explains what and why. If using CodeRabbit's summary, please make sure it makes sense.
PR Follows TRT-LLM CODING GUIDELINES to the best of your knowledge.
Test cases are provided for new code paths (see test instructions)
Any new dependencies have been scanned for license and vulnerabilities
CODEOWNERS updated if ownership changes
Documentation updated as needed
The reviewers assigned automatically/manually are appropriate for the PR.
Please check this after reviewing the above items as appropriate for this PR.

GitHub Bot Help

/bot [-h] ['run', 'kill', 'skip', 'reuse-pipeline'] ...

Provide a user friendly way for developers to interact with a Jenkins server.

Run /bot [-h|--help] to print this help message.

See details below for each supported subcommand.

run [--reuse-test (optional)pipeline-id --disable-fail-fast --skip-test --stage-list "A10-PyTorch-1, xxx" --gpu-type "A30, H100_PCIe" --test-backend "pytorch, cpp" --add-multi-gpu-test --only-multi-gpu-test --disable-multi-gpu-test --post-merge --extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx" --detailed-log --debug(experimental)]

Launch build/test pipelines. All previously running jobs will be killed.

--reuse-test (optional)pipeline-id (OPTIONAL) : Allow the new pipeline to reuse build artifacts and skip successful test stages from a specified pipeline or the last pipeline if no pipeline-id is indicated. If the Git commit ID has changed, this option will be always ignored. The DEFAULT behavior of the bot is to reuse build artifacts and successful test results from the last pipeline.

--disable-reuse-test (OPTIONAL) : Explicitly prevent the pipeline from reusing build artifacts and skipping successful test stages from a previous pipeline. Ensure that all builds and tests are run regardless of previous successes.

--disable-fail-fast (OPTIONAL) : Disable fail fast on build/tests/infra failures.

--skip-test (OPTIONAL) : Skip all test stages, but still run build stages, package stages and sanity check stages. Note: Does NOT update GitHub check status.

--stage-list "A10-PyTorch-1, xxx" (OPTIONAL) : Only run the specified test stages. Examples: "A10-PyTorch-1, xxx". Note: Does NOT update GitHub check status.

--gpu-type "A30, H100_PCIe" (OPTIONAL) : Only run the test stages on the specified GPU types. Examples: "A30, H100_PCIe". Note: Does NOT update GitHub check status.

--test-backend "pytorch, cpp" (OPTIONAL) : Skip test stages which don't match the specified backends. Only support [pytorch, cpp, tensorrt, triton]. Examples: "pytorch, cpp" (does not run test stages with tensorrt or triton backend). Note: Does NOT update GitHub pipeline status.

--only-multi-gpu-test (OPTIONAL) : Only run the multi-GPU tests. Note: Does NOT update GitHub check status.

--disable-multi-gpu-test (OPTIONAL) : Disable the multi-GPU tests. Note: Does NOT update GitHub check status.

--add-multi-gpu-test (OPTIONAL) : Force run the multi-GPU tests in addition to running L0 pre-merge pipeline.

--post-merge (OPTIONAL) : Run the L0 post-merge pipeline instead of the ordinary L0 pre-merge pipeline.

--extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx" (OPTIONAL) : Run the ordinary L0 pre-merge pipeline and specified test stages. Examples: --extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx".

--detailed-log (OPTIONAL) : Enable flushing out all logs to the Jenkins console. This will significantly increase the log volume and may slow down the job.

--debug (OPTIONAL) : Experimental feature. Enable access to the CI container for debugging purpose. Note: Specify exactly one stage in the stage-list parameter to access the appropriate container environment. Note: Does NOT update GitHub check status.

For guidance on mapping tests to stage names, see docs/source/reference/ci-overview.md
and the scripts/test_to_stage_mapping.py helper.

kill

kill

Kill all running builds associated with pull request.

skip

skip --comment COMMENT

Skip testing for latest commit on pull request. --comment "Reason for skipping build/test" is required. IMPORTANT NOTE: This is dangerous since lack of user care and validation can cause top of tree to break.

reuse-pipeline

reuse-pipeline

Reuse a previous pipeline to validate current commit. This action will also kill all currently running builds associated with the pull request. IMPORTANT NOTE: This is dangerous since lack of user care and validation can cause top of tree to break.

yilin-void · 2025-09-28T03:30:42Z

/bot run

coderabbitai · 2025-09-28T03:31:20Z

📝 Walkthrough

Walkthrough

Introduces a helper method to determine post-quant alltoall support and replaces inline conditions with this method in forward_chunk. Adds runtime assertions enforcing nvfp4 when using DeepEP post-quant alltoall. Adjusts control flow for post-quant alltoall selection without changing external interfaces.

Changes

Cohort / File(s)	Summary
MoE WideEP post-quant alltoall control flow `tensorrt_llm/_torch/modules/fused_moe/fused_moe_wide_ep.py`	Added WideEPMoE.is_post_quant_all2all_supported(); replaced composite condition in forward_chunk with helper; inserted assertions requiring nvfp4 for DeepEP post-quant alltoall; centralized decision logic for MNNVL, DeepEP, DeepEPLowLatency paths.

Sequence Diagram(s)

sequenceDiagram
  participant Caller
  participant WideEPMoE
  participant AllToAll

  Caller->>WideEPMoE: forward_chunk(...)
  activate WideEPMoE
  WideEPMoE->>WideEPMoE: is_post_quant_all2all_supported()
  alt post-quant alltoall supported
    note over WideEPMoE: DeepEP requires nvfp4 (asserts)
    WideEPMoE->>AllToAll: post-quant alltoall dispatch
    AllToAll-->>WideEPMoE: tensors
  else not supported
    WideEPMoE->>AllToAll: standard path (reduce-scatter/all-reduce as applicable)
    AllToAll-->>WideEPMoE: tensors
  end
  WideEPMoE-->>Caller: outputs
  deactivate WideEPMoE

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~25 minutes

Pre-merge checks and finishing touches

❌ Failed checks (3 warnings)

Check name	Status	Explanation	Resolution
Title Check	⚠️ Warning	The title follows the prescribed format but narrowly describes only the DeepEP nvfp4 restriction and omits the main change of centralizing post-quantization all2all support logic across multiple backend types via a new helper method.	Please update the title to clearly reflect the centralization of post-quantization all2all support logic and the specific restrictions applied to DeepEP and DeepEPLowLatency, for example “Consolidate post-quantization all2all support and restrict DeepEP to nvfp4.”
Description Check	⚠️ Warning	The PR description is still the default template with empty Description and Test Coverage sections and contains no summary of the issue, implementation details, or tests, making it impossible to understand the changes or verify coverage.	Please fill in the Description section with a clear overview of the problem and the implemented solution, list relevant test cases in the Test Coverage section, and replace all template placeholders with actual details to make the PR reviewable.
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	You can run `@coderabbitai generate docstrings` to improve docstring coverage.

✨ Finishing touches

📝 Generate Docstrings

🧪 Generate unit tests

Create PR with unit tests
Post copyable unit tests in a comment

📜 Recent review details

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 77b68d9 and ffa9977.

📒 Files selected for processing (1)

tensorrt_llm/_torch/modules/fused_moe/fused_moe_wide_ep.py (3 hunks)

🧰 Additional context used

📓 Path-based instructions (3)

**/*.{h,hpp,hh,hxx,cpp,cxx,cc,cu,cuh,py}

📄 CodeRabbit inference engine (CODING_GUIDELINES.md)

Use only spaces, no tabs; indent with 4 spaces.

Files:

tensorrt_llm/_torch/modules/fused_moe/fused_moe_wide_ep.py

**/*.py

📄 CodeRabbit inference engine (CODING_GUIDELINES.md)

**/*.py: Python code must target Python 3.8+.
Indent Python code with 4 spaces; do not use tabs.
Maintain module namespace when importing; prefer 'from package.subpackage import foo' then 'foo.SomeClass()' instead of importing the class directly.
Python filenames should be snake_case (e.g., some_file.py).
Python classes use PascalCase names.
Functions and methods use snake_case names.
Local variables use snake_case; prefix 'k' for variables that start with a number (e.g., k_99th_percentile).
Global variables use upper SNAKE_CASE prefixed with 'G' (e.g., G_MY_GLOBAL).
Constants use upper SNAKE_CASE (e.g., MY_CONSTANT).
Avoid shadowing variables from an outer scope.
Initialize all externally visible members of a class in the constructor.
Prefer docstrings for interfaces that may be used outside a file; comments for in-function or file-local interfaces.
Use Google-style docstrings for classes and functions (Sphinx-parsable).
Document attributes and variables inline so they render under the class/function docstring.
Avoid reflection when a simpler, explicit approach suffices (e.g., avoid dict(**locals()) patterns).
In try/except, catch the most specific exceptions possible.
For duck-typing try/except, keep the try body minimal and use else for the main logic.

Files:

tensorrt_llm/_torch/modules/fused_moe/fused_moe_wide_ep.py

**/*.{cpp,cxx,cc,h,hpp,hh,hxx,cu,cuh,py}

📄 CodeRabbit inference engine (CODING_GUIDELINES.md)

Prepend the NVIDIA Apache-2.0 copyright header with current year to the top of all source files (e.g., .cpp, .h, .cu, .py).

Files:

tensorrt_llm/_torch/modules/fused_moe/fused_moe_wide_ep.py

🧬 Code graph analysis (1)

tensorrt_llm/_torch/modules/fused_moe/fused_moe_wide_ep.py (2)

tensorrt_llm/_torch/modules/fused_moe/interface.py (2)

has_nvfp4 (299-302)

has_fp8_qdq (287-290)

tensorrt_llm/_torch/modules/fused_moe/fused_moe_cutlass.py (1)

has_w4afp8 (181-184)

🔇 Additional comments (3)

tensorrt_llm/_torch/modules/fused_moe/fused_moe_wide_ep.py (3)

421-431: Helper consolidates the gating logic cleanly.

Centralizing the post-quant all2all support checks here makes the allowed combinations explicit and guarantees the DeepEP+nvfp4 restriction can’t be bypassed elsewhere.

508-510: Forward path now defers to the helper as intended.

This keeps the runtime decision aligned with the consolidated logic and avoids duplicating the capability matrix in forward_chunk.

630-631: Assert fortifies the DeepEP post-quant guardrail.

The explicit nvfp4 assertion ensures we fail fast if the helper’s contract is ever violated, preventing unsupported execution.

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

tensorrt-cicd · 2025-09-28T03:35:43Z

PR_Github #20155 [ run ] triggered by Bot

tensorrt-cicd · 2025-09-28T06:39:32Z

PR_Github #20155 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #15195 completed with status: 'FAILURE'

Signed-off-by: Yilin Zhang <18275976+yilin-void@users.noreply.github.com>

yilin-void · 2025-09-29T01:35:44Z

/bot run

tensorrt-cicd · 2025-09-29T01:41:22Z

PR_Github #20193 [ run ] triggered by Bot

tensorrt-cicd · 2025-09-29T06:12:32Z

PR_Github #20193 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #15229 completed with status: 'SUCCESS'

yilin-void requested a review from a team as a code owner September 28, 2025 03:29

yilin-void requested a review from yuxianq September 28, 2025 03:29

yilin-void force-pushed the fix branch from 8400f07 to ffa9977 Compare September 28, 2025 03:30

yilin-void requested a review from yifeizhang-c September 28, 2025 03:30

yifeizhang-c approved these changes Sep 29, 2025

View reviewed changes

only support deepep post quant all2all on nvfp4

94a7c4a

Signed-off-by: Yilin Zhang <18275976+yilin-void@users.noreply.github.com>

yilin-void force-pushed the fix branch from ffa9977 to 94a7c4a Compare September 29, 2025 01:34

yuxianq approved these changes Sep 29, 2025

View reviewed changes

yilin-void merged commit 7f1e2db into NVIDIA:main Sep 29, 2025
5 checks passed

kaiyux mentioned this pull request Oct 14, 2025

[None][fix] Fix is_post_quant_all2all_supported for MNNVL #8355

Merged

1 task

yilin-void deleted the fix branch October 20, 2025 13:39

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[None][fix] only support deepep post quant all2all on nvfp4 #8041

[None][fix] only support deepep post quant all2all on nvfp4 #8041

Uh oh!

yilin-void commented Sep 28, 2025 •

edited by coderabbitai bot

Loading

Uh oh!

yilin-void commented Sep 28, 2025

Uh oh!

coderabbitai bot commented Sep 28, 2025 •

edited

Loading

Walkthrough

Changes

Sequence Diagram(s)

Estimated code review effort

Uh oh!

tensorrt-cicd commented Sep 28, 2025

Uh oh!

tensorrt-cicd commented Sep 28, 2025

Uh oh!

yilin-void commented Sep 29, 2025

Uh oh!

tensorrt-cicd commented Sep 29, 2025

Uh oh!

tensorrt-cicd commented Sep 29, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

[None][fix] only support deepep post quant all2all on nvfp4 #8041

[None][fix] only support deepep post quant all2all on nvfp4 #8041

Uh oh!

Conversation

yilin-void commented Sep 28, 2025 • edited by coderabbitai bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary by CodeRabbit

Description

Test Coverage

PR Checklist

GitHub Bot Help

kill

skip

reuse-pipeline

Uh oh!

yilin-void commented Sep 28, 2025

Uh oh!

coderabbitai bot commented Sep 28, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Walkthrough

Changes

Sequence Diagram(s)

Estimated code review effort

Pre-merge checks and finishing touches

Uh oh!

tensorrt-cicd commented Sep 28, 2025

Uh oh!

tensorrt-cicd commented Sep 28, 2025

Uh oh!

yilin-void commented Sep 29, 2025

Uh oh!

tensorrt-cicd commented Sep 29, 2025

Uh oh!

tensorrt-cicd commented Sep 29, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

yilin-void commented Sep 28, 2025 •

edited by coderabbitai bot

Loading

coderabbitai bot commented Sep 28, 2025 •

edited

Loading